Beating the AI Doom Cycle

What It Covers

Host Nathaniel Whittemore maps a five-stage "AI Doom Cycle" — from skepticism through AI psychosis, doom desperation, and real-world recalibration — arguing that reaching the final stage of "enlightened excitement" enables more productive policy conversations and practical AI adoption strategies.

Key Questions Answered

• •The AI Doom Cycle framework: The five stages — skepticism, AI psychosis, doom desperation, real-world recalibration, and enlightened excitement — describe how individuals emotionally process AI's rise. Most people currently oscillate between stages two and three simultaneously. Recognizing which stage you occupy helps calibrate responses to AI news and avoid reactive decision-making based on extreme narratives.
• •Compute scarcity is slowing displacement timelines: Anthropic shifted enterprise Claude Code pricing from a $200 flat-rate to usage-based billing, with GitHub following. GitHub Copilot users discovered their subsidized $451 plans would cost $11,432 under actual usage pricing — a 25x gap. This structural compute shortage means mass workforce automation is physically and economically constrained far beyond what doom narratives suggest.
• •Citadel's 15–25% productivity benchmark: CEO Ken Griffin reported AI agents completing weeks or months of PhD-level finance research in hours or days, yielding a 15–25% firm-wide productivity boost. Rather than inspiring optimism, this finding made Griffin depressed about societal impact — illustrating how increased AI capability belief reliably pushes individuals from AI psychosis directly into doom desperation.
• •The relational economy as a hedge: Economist Alex Imas' essay "What Will Be Scarce" argues that sectors where human provenance is part of the economic value — craftsmanship, personal services, human-created art — will grow proportionally as AI commoditizes output-based work. Positioning skills and career investments toward relational value creation offers a concrete strategy for navigating AI-driven labor market shifts.
• •Policy windows exist now, before infrastructure locks in: Local communities and policymakers currently hold leverage over data center permitting and operating conditions. Proposals circulating include federally taxing tokens at under $0.50 per million at the provider level — potentially generating $10 billion annually — and mandating affordable GPU set-asides for low-income access. Acting during the build-out phase captures options unavailable post-construction.